1. Field of the Invention
The present invention relates to methods and nucleic acid vector compositions for modifying gene expressing, involving the preparation and use of improved retroviral vectors which encode antisense RNA molecules or, alternatively, transcriptionally active RNAs that encode selected proteins. The retroviral constructs of the present invention may be employed for introducing desired gene expression units into selected target cells, such as into tumor cells within individuals afflicted with cancer.
2. Description of the Related Art
It is now well established that a variety of diseases, ranging from certain cancers to various genetic defects, are caused, at least in part, by genetic abnormalities that result in either the over expression of one or more genes, or the expression of an abnormal or mutant gene or genes. For example, many forms of cancer in man are now known to be the result of, at least indirectly, the expression of "oncogenes". Oncogenes are genetically altered genes whose altered expression product somehow disrupts normal cellular function or control (Spandidos, et al., 1989).
Most oncogenes studied to date have been found to be "activated" as the result of a mutation, often a point mutation, in the coding region of a normal cellular gene or of a "protooncogene", that results in amino acid substitutions in the protein expression product. This altered expression product, in turn, exhibits an abnormal biological function that somehow takes part in the neoplastic process (Travali, et al., 1990). The underlying mutations can arise by various means, such as by chemical mutagenesis or ionizing radiation.
A number of oncogenes have now been identified and characterized to varying degrees, including ras, myc, neu, raf, erb, src, fms, jun and abl (Travali, et al., 1990; Minna, 1989; Bishop, 1987). It is likely that as our knowledge of tumorigenesis increases, additional oncogenes will be identified and characterized. Many of the foregoing, including ras, myc and erbB, comprise families of genes, whose expression product bear sequence similarities to other members of the family (Shih, et al., 1984; Bos, 1989; Schwab, et al., 1985). In the case of many of these gene families, it is typical that oncogenesis involves an activation of only one member of the family, with other "unactivated" members serving a role in normal cellular functions (Id.).
The study of DNA-mediated gene transfer has revealed the existence of activated cellular oncogenes in a variety of human tumors (for review, see Cooper, et al., 1982). Oncogenes have been identified in human bladder, colon, lung and mammary carcinoma cell lines (Krontiris, et al., 1981; Murray, et al., 1981; Perucho, et al., 1981), promyelocytic leukemia (Murray, et al., 1981), neuroblastoma (Shimizu, et al., 1983) and sarcoma cell lines (Pulciani, et al., 1982), and various solid tumors including carcinomas of the lung, and pancreas (Pulciani, et al., 1982). Studies have demonstrated that various transforming genes detected by transfection correspond to activated cellular homologues of retroviral oncogenes (Pulciani, et al., 1982; Der, et al., 1982; Parada, et al., 1982; Santos, et al., 1982), although others have no known retroviral cognate (Tulciani, et al., 1982; Lane, et al., 1982).
The ras oncogene family has been perhaps the best characterized to date (Barbacid, 1987; Bos, 1989). Most of the identified transforming genes in human carcinomas have been a member of the ras gene family, which encode immunologically related proteins having a molecular weight of 21,000 (p21) (Ellis, et al., 1981; Papageorge, et al., 1982). This family is comprised of at least 3 members, one transduces as H-ras in the Harvey strain of murine sarcoma virus (Ellis, et al., 1981), one as K-ras and Kirsten murine sarcoma virus (Ellis, et al., 1981), and one identified by low stringency hybridization to H-ras, termed N-ras (Shimizu, et al., 1983). As noted, all members of the ras gene family encode closely related proteins of approximately 21,000 Daltons which have been designated p21s (Ellis, et al, 1981). The level of p21 expression is similar in many different human tumor cells, independent of whether the cell contains an activated ras gene detectable by transfection.
Nucleotide sequence analysis of the H-ras transforming gene of the EJ human bladder carcinoma has indicated that the transforming activity of this gene is a consequence of a point mutation altering amino acid 12 of p21 from glycine to valine (Tabin, et al., 1982). Studies of proteins encoded by K-ras genes activated in four human lung and colon carcinoma cell lines indicated that the transforming activity of K-ras in these human tumors was also a consequence of structural mutations (Der and Cooper, 1983). Other mutations have been found to result in ras gene activation as well. For example, the H-ras gene activated in a lung carcinoma cell line encodes the normal amino acid position 12 but is mutated at codon 61 to encode leucine rather than glutamine (Yuasa, et al., 1983). An N-ras gene activated in a human neuroblastoma cell line is also mutated at codon 61 but encodes lysine rather that glutamine (Taparowski, et al. 1983). Thus, studies such as these have indicated that ras genes in human neoplasms are commonly activated by structural mutations, often point mutations, that thus far occur at codon 12 or 61 with different amino acid substitutions resulting in ras gene activation in different tumors.
Antisense RNA technology has been developed as one approach to inhibiting gene expression, particularly oncogene expression. An "antisense" RNA molecule is one which contains the complement of, and can therefore hybridize with, protein-encoding RNAs of the cell. It is believed that the hybridization of antisense RNA to its cellular RNA complement can prevent expression of the cellular RNA, perhaps by limiting its translatability. While various studies have involved the processing of RNA or direct introduction of antisense RNA oligonucleotides to cells for the inhibition of gene expression (Brown, et al., 1989; Wickstrom, et al., 1988; Smith, et al., 1986; Buvoli, et al., 1987), the more common means of cellular introduction of antisense RNAs has been through the construction of recombinant vectors which will express antisense RNA once the vector is introduced into the cell.
A principal application of antisense RNA technology has been in connection with attempts to affect the expression of specific genes. For example, Delauney, et al. have reported the use antisense transcripts to inhibit gene expression in transgenic plants (Delauney, et al., 1988). These authors report the down-regulation of chloramphenicol acetyl transferase activity in tobacco plants transformed with CAT sequences through the application of antisense technology.
Antisense technology has also been applied in attempts to inhibit the expression of various oncogenes. For example, Kasid, et al., 1989, report the preparation of recombinant vector construct employing Craf-1 cDNA fragments in an antisense orientation, brought under the control of an adenovirus 2 late promoter. These authors report that the introduction of this recombinant construct into a human squamous carcinoma resulted in a greatly reduced tumorigenic potential relative to cells transfected with control sense transfectants. Similarly, Prochownik, et al., 1988, have reported the use of Cmyc antisense constructs to accelerate differentiation and inhibit G.sub.1 progression in Friend Murine Erythroleukemia cells. In contrast, Khokha, et al., 1989, discloses the use of antisense RNAs to confer oncogenicity on 3T3 cells, through the use of antisense RNA to reduce murine tissue inhibitor or metalloproteinases levels.
Unfortunately, the use of current antisense technology often results in failure, particularly where one seeks to selectively inhibit a member of a gene family. One reason for this failure can be traced to the high expression levels of antisense message that are apparently required for inhibition. Unfortunately, the requisite expression levels of antisense message has not been generally achievable with existing constructs. Problems have also arisen due to the similarity in underlying DNA sequences, which results in the cross-hybridization of antisense RNA, retarding the expression of genes required for normal cellular functions. An example is presented by Debus, et al., 1990, who reported that in the case of ras oncogenes, antisense ras oligonucleotides kill both normal and cancer cells, which, of course, is not a desired effect.
Another important "oncogene" is the gene encoding the p53 cellular protein. The p53 gene is one of the most common targets for genetic abnormalities in human tumors (Hollstein et al., 1991). For example, it has been reported that p53 mutations occur in all histological types of lung cancer at frequencies of about 75% in small cell lung cancer (SCLC) and about 50% in non small cell lung cancer (NSCLC) (Takahashi et al., 1991). Evidence suggests that p53 acts as a "tumor suppressor" gene, and its inactivation through mutation can lead to oncogenic development. In fact, a predominance of G to T transversions in p53 and ras mutations in lung cancer, as well as epidemiological data, supports a close association between smoking and p53 mutations in NSCLC have suggested that p53 is a candidate for molecular targets of genetic damage caused by cigarette smoke (Zakut-Houri et al., 1985).
One approach that has been suggested as a means of treatment of such tumors is the introduction of so-called "wild-type" or non-mutated p53 (wt-p53) into affected cells, e.g., through the use of retroviral vectors which encode the wild type protein (Takahashi et al., 1992; Lee et al., EP appl. publ. 0 475 623 A1). The vectors proposed by these individuals included a wt-p53 genes wherein the direction of transcription of the encoded wt-p53 was in the same orientation as that of the retroviral long terminal repeats (LTRs). Unfortunately, in studies conducted by the present inventors reported hereinbelow, the ability of retroviral wt-p53 constructs prepared having such an orientation to suppress tumor growth was found to be less than optimal. Presumably, this shortcoming is the result of poor expression of the wt-p53 gene in the target cells.
Therefore, while it is clear that current gene therapy technology shows potential promise as a means of external control of gene expression, it is equally clear that it does suffer particular draw backs, such as the need for high level expression and a lack of selectivity where gene families are concerned. There is a particular need, therefore, for a general approach to the design of gene therapy protocols that will allow selective inhibition of gene expression, even in the case of closely related genes.